Development of a "best representativity" method for experimental data analysis and an application to the critical experiments at the Toshiba NCA facility

For nuclear critical experiments, it is essential to certify similarities of the experiment with the objective of the actual reactor conditions or actual reactor equipment. To judge the applicability of the experimental data, the concept of a "representativity factor" has recently been adopted in the critical experiment field, particularly for fast breeder reactors and future reactor studies. In this study, we extended this concept to the design of a light water reactor system. We developed a new numerical evaluation method and a calculation system. The method is based on a linear combination of the sensitivity coefficient vector of an experiment in which the representativity factor to the target system is maximized to utilize experimental data effectively. Simultaneously, using the measurement data of critical experiments, the method enables us to evaluate calculation errors caused by errors or uncertainties of physical parameters. The derivation of the new calculation method is explained first. We then qualify it with a sample calculation, presenting numerical results for three kinds of critical experiments conducted at the Toshiba Nuclear Critical Assembly facility. Finally, the results are compared with those of an extended bias factor method to clarify the performance of the new method. (author)